AI Sovereignty Risks Put Governance and Visibility in Focus

Businesses may lose access to artificial intelligence tools with little warning as government decisions, regulation and provider policies change. A June dispute involving Anthropic illustrates how quickly that can happen, according to TechRadar Pro Perspectives contributor Jon Abbott.

In June, Anthropic released two of its most capable AI models. Within days, the US government ordered the company to cut off access for users outside the country, citing national security concerns. Anthropic could not reliably verify where users were logging in from, so it withdrew both models for everyone, including US customers. Two and a half weeks later, access was restored. One model remains limited to a small number of approved US organizations.

Abbott says the episode shows why companies should prepare for changes to AI access, functionality and governance rather than try to predict the next disruption. He recommends mapping where AI is used, which teams and processes depend on each tool, and having a contingency plan. Companies also need to understand what data each tool can access and who can use it.

The article says AI capabilities are concentrated among a small number of providers, while export controls for frontier AI are still taking shape. It also describes an agreement between Anthropic and the US government on flagging and reviewing future risks. Abbott argues that businesses need visibility into their AI use and tested continuity plans, while independent oversight of model safety should be strengthened.